Anyone else’s head spinning trying to make sense of the AI industry right now? There are key “tug-of-wars” happening in the AI industry that deserve calling out.
Making sense of AI’s contradictions
First up: Is AI fueling the economy while simultaneously crushing the job market by slowing hiring? Or is it unproductive with no measurable ROI? Both outcomes may be true, and they are not mutually exclusive.
- The AI adoption bull case: Derek Thompson’s breakdown and interview with Stanford researchers shows how recent job data suggests AI’s impact on hiring is “plausibly yes” a significant factor.
- The AI adoption bear case: Mike Judge’s take that even where AI is being adopted the most (software), AI is not actually making us more productive. He points to the lack of “shovelware,” a low-quality but easy-to-shove-out software.
AI labs split over chip strategy
Another rift centers on hardware: to rely on Nvidia or not?
- OpenAI is set to begin mass production of its own AI chips with Broadcom in a $10B deal.
- Google has expanded from using its chips solely for internal training to selling them to other cloud providers.
- Amazon requires Anthropic, which recently raised $13B, to train new Claude models on its Trainium chips. SemiAnalysis also published an in-depth analysis of the deal.
These moves are expected to reduce reliance on Nvidia. The company’s chips remain expensive, and its $4.6 trillion market cap saw its stock dip on the OpenAI-Broadcom news.
The debate over evals
The newest rift: to use evals, or not?
AI “evals” are quality control mechanisms that ensure AI systems function correctly and fairly before influencing real-world decisions such as job applications, medical diagnoses, or loan approvals. “Evals” are akin to car safety-test cars or restaurant inspections.
The AI industry is pretty split on whether or not these evals actually matter.
- Swyx summarized the debate well in a single tweet, pointing out that major labs eval based on “vibes” and not rigid testing, while Julia Neagu dove deep into the topic if you’re actually considering building with AI.
- This is important, as everyone these days is trying to spin up what’s called “RL environments,” which are places where AI can run around and learn things via reinforcement learning, like Prime Intellect / Mechanize.
Shreya Shankar wrote this great piece in defense of evals, arguing that teams claiming they don’t need systematic evaluation are usually already doing it through dogfooding and error analysis — they just don’t call it “evals” — and that dismissing formal evaluation frameworks particularly harms new AI builders who lack the domain expertise to rely on intuition alone.
Allie K. Miller added her own retort in favor of evals, outlining 6 reasons why you actually really do want evals: Is this AI good enough for job augmentation and/or replacement? Does it create business value? Does it help us defeat hype? Can the AI actually complete a goal? And are the AI labs actually improving their AI or just changing things?.
Open source vs. closed source
Then there’s the age old (well, in AI years, anyway) philosophical split: to go open source or not?
While American AI labs are locking down their top models tighter than Fort Knox, Chinese companies just dropped two unique trillion-parameter models (that’s trillions of numbers determining AI responses, btw!) in one weekend:
- Alibaba’s Qwen3-Max-Preview
- Moonshot’s Kimi K2-Instruct
Both claim major performance increases (read more about them here).
What gives?? In a nutshell: Chinese labs appear to be moving fast and breaking things while U.S. labs are moving slow(er) and fixing things (like GPT-5).
Paging Google DeepMind… we need Gemini 3 Pro, pronto! We need a Bat signal equivalent for Logan Kilpatrick and Demis Hassabis… maybe a giant (nano) banana??
Now, the new Qwen Max is actually closed behind their API atm… their first non-open-model (that we know of, anyway).
As the adage goes: Open source when you’re behind, closed source when you’re ahead.
So are Chinese labs feeling confident they’re gaining on US peers? Or is this just the reality of unleashing something so big: ya kinda need to actually monetize it??
Our take
This is all probably signs of a healthy and growing ecosystem. There’s lots of debate, there’s lots of opportunity, there’s lots to still be learned. As Mike Knoop of ARC-AGI says, there’s only really been two major breakthroughs in language models: the original transformer discovery (tech that powers ChatGPT) and chain of thought reasoning (the “thinking” mode). Translation? We’re still early, ppl.
Editor’s note: This content originally ran in our sister publication, The Neuron. To read more from The Neuron, sign up for its newsletter here.